xinyin025/one-api
1
0
Fork 0
mirror of https://github.com/songquanpeng/one-api.git synced 2025-09-17 17:16:38 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

feat: enhance OpenAI adaptor to support new models and enforce usage in streaming

Browse Source
This commit is contained in:
Laisky.Cai 2024-11-27 06:43:30 +00:00
parent 6ab87f8a08
commit 22de24de4c
4 changed files with 42 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
relay/adaptor/openai/adaptor.go
View File

@ -8,6 +8,7 @@ import (
"strings"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common/config"
"github.com/songquanpeng/one-api/relay/adaptor"
"github.com/songquanpeng/one-api/relay/adaptor/doubao"
"github.com/songquanpeng/one-api/relay/adaptor/minimax"
@ -75,13 +76,29 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
if request == nil {
return nil, errors.New("request is nil")
}
if request.Stream {
if config.EnforceIncludeUsage && request.Stream {
// always return usage in stream mode
if request.StreamOptions == nil {
request.StreamOptions = &model.StreamOptions{}
}
request.StreamOptions.IncludeUsage = true
}
// o1 do not support system prompt and max_tokens
if strings.HasPrefix(request.Model, "o1-mini") || strings.HasPrefix(request.Model, "o1-preview") {
request.MaxTokens = 0
request.Messages = func(raw []model.Message) (filtered []model.Message) {
for i := range raw {
if raw[i].Role != "system" {
filtered = append(filtered, raw[i])
}
}
return
}(request.Messages)
}
return request, nil
}

6
relay/adaptor/openai/constants.go
View File

@ -7,9 +7,7 @@ var ModelList = []string{
"gpt-4", "gpt-4-0314", "gpt-4-0613", "gpt-4-1106-preview", "gpt-4-0125-preview",
"gpt-4-32k", "gpt-4-32k-0314", "gpt-4-32k-0613",
"gpt-4-turbo-preview", "gpt-4-turbo", "gpt-4-turbo-2024-04-09",
"gpt-4o", "gpt-4o-2024-05-13",
"gpt-4o-2024-08-06",
"chatgpt-4o-latest",
"gpt-4o", "gpt-4o-2024-05-13", "gpt-4o-2024-08-06", "gpt-4o-2024-11-20", "chatgpt-4o-latest",
"gpt-4o-mini", "gpt-4o-mini-2024-07-18",
"gpt-4-vision-preview",
"text-embedding-ada-002", "text-embedding-3-small", "text-embedding-3-large",
@ -20,4 +18,6 @@ var ModelList = []string{
"dall-e-2", "dall-e-3",
"whisper-1",
"tts-1", "tts-1-1106", "tts-1-hd", "tts-1-hd-1106",
"o1-preview", "o1-preview-2024-09-12",
"o1-mini", "o1-mini-2024-09-12",
}

11
relay/billing/ratio/model.go
View File

@ -37,6 +37,7 @@ var ModelRatio = map[string]float64{
"chatgpt-4o-latest": 2.5, // $0.005 / 1K tokens
"gpt-4o-2024-05-13": 2.5, // $0.005 / 1K tokens
"gpt-4o-2024-08-06": 1.25, // $0.0025 / 1K tokens
"gpt-4o-2024-11-20": 1.25, // $0.0025 / 1K tokens
"gpt-4o-mini": 0.075, // $0.00015 / 1K tokens
"gpt-4o-mini-2024-07-18": 0.075, // $0.00015 / 1K tokens
"gpt-4-vision-preview": 5, // $0.01 / 1K tokens
@ -48,6 +49,10 @@ var ModelRatio = map[string]float64{
"gpt-3.5-turbo-instruct": 0.75, // $0.0015 / 1K tokens
"gpt-3.5-turbo-1106": 0.5, // $0.001 / 1K tokens
"gpt-3.5-turbo-0125": 0.25, // $0.0005 / 1K tokens
"o1-preview": 7.5, // $15.00 / 1M input tokens
"o1-preview-2024-09-12": 7.5,
"o1-mini": 1.5, // $3.00 / 1M input tokens
"o1-mini-2024-09-12": 1.5,
"davinci-002": 1, // $0.002 / 1K tokens
"babbage-002": 0.2, // $0.0004 / 1K tokens
"text-ada-001": 0.2,
@ -344,6 +349,12 @@ func GetCompletionRatio(name string, channelType int) float64 {
}
return 2
}
if strings.HasPrefix(name, "o1-preview") {
return 4
}
if strings.HasPrefix(name, "o1-mini") {
return 4
}
if name == "chatgpt-4o-latest" {
return 3
}

12
relay/controller/text.go
View File

@ -4,11 +4,11 @@ import (
"bytes"
"encoding/json"
"fmt"
"github.com/songquanpeng/one-api/common/config"
"io"
"net/http"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common/config"
"github.com/songquanpeng/one-api/common/logger"
"github.com/songquanpeng/one-api/relay"
"github.com/songquanpeng/one-api/relay/adaptor"
@ -19,6 +19,7 @@ import (
"github.com/songquanpeng/one-api/relay/channeltype"
"github.com/songquanpeng/one-api/relay/meta"
"github.com/songquanpeng/one-api/relay/model"
relaymodel "github.com/songquanpeng/one-api/relay/model"
)
func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
@ -86,9 +87,12 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
return nil
}
func getRequestBody(c *gin.Context, meta *meta.Meta, textRequest *model.GeneralOpenAIRequest, adaptor adaptor.Adaptor) (io.Reader, error) {
if !config.EnforceIncludeUsage && meta.APIType == apitype.OpenAI && meta.OriginModelName == meta.ActualModelName && meta.ChannelType != channeltype.Baichuan {
// no need to convert request for openai
func getRequestBody(c *gin.Context, meta *meta.Meta, textRequest *relaymodel.GeneralOpenAIRequest, adaptor adaptor.Adaptor) (io.Reader, error) {
if !config.EnforceIncludeUsage &&
meta.APIType == apitype.OpenAI &&
meta.OriginModelName == meta.ActualModelName &&
meta.ChannelType != channeltype.OpenAI && // openai also need to convert request
meta.ChannelType != channeltype.Baichuan {
return c.Request.Body, nil
}